ML20086G521

Technical Evaluation Rept on Individual Plant Exam Back-End Analysis

ML20086G521
Person / Time
Site:	Hatch
Issue date:	12/31/1994
From:	Budnitz R SCIENCE & ENGINEERING ASSOCIATES, INC.
To:	NRC
Shared Package
ML20086G518	List: ML20086G516 ML20086G521 ML20086G528
References
CON-NRC-05-91-068, CON-NRC-5-91-68 SCIE-NRC-228-94, NUDOCS 9507140442
Download: ML20086G521 (25)

Text

.-

'd- 4 SCIE-NRC-228-94 EDWIN I. HATCH NUCLEAR PLANT (UNITS 1 AND 2)

TECHNICAL EVALUATION REPORT >

ON THE INDIVIDUAL PLANT EXAMINATION BACK-END ANALYSIS Robert J. Budnitz Prepared for the U.S. Nuclear Regulatory Commission Under Contract NRC-05-91-068-27 December 1994 SCIENTECH, Inc. ,

11140 Rockville Pike Rockville, Maryland 20852 '

q

% mowt

?

t TABLE OF CONTENTS EXECUTIVE

SUMMARY

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii 1  !

1. INTRODUCTION .....................................

CONTRACTOR REVIEW FINDINGS ........................ 2 2.

Review and Identification of IPE Insights ' ........................ 2 2.1 ,

2.1.1 General Review of IPE Back-End Analytical Process ............. 2

• 2.1.1.1 Completeness ........................... 2 2.1.1.2 Description, Justification, and Consistency ......... 2 2.1.1.3 Process Used for IPE ...................... 2 '

2.1.1.4 Peer Review of IPE ....................... 2 ,

2.1.2 Containment Analysis / Characterization ..................... 3 2.1.2.1 Front-end Back-end Dependencies ............... 3 2.1.2.2 Sequences with Significant Probability ............ 4 2.1.2.3 Failure Modes and Timing ................... 4 2.1.2.4 Containment Isolation Failure ................. 5 2.1.2.5 System / Human Responses ....,............... 6 2.1.2.6 Radionuclide Release Characterization ............ 6 2.1.3 Accident Progression and Containment Performance Analysis ....... 7 2.1.3.1 Severe Accident Progression .................. 7 2.1.3.2 Dominant Contributors: Consistency with IPE Insights .. 7 2.1.3.3 Characterization of Containment Performance ....... 8 2.1.3.4 Impact on Equipment Behavior ................ 8 2.1.4 Reducing Probability of Core Damage or Fission Product Release .... 9 l 2.1.4.1 Definition of Vulnerability ................... 9 l 2.1.4.2 Plant Improvements ....................... 9 2.1.5 Responses to CPI Program Recommendations . . . . . . . . . . . . . . . 10 1

2.2 IPE Strengths and Weaknesses .............................10 2.2.1 IPE Strengths ...................................10 2.2.2 IPE Weaknesses ..................................11

3. OVERALL EVALUATION ................................11

4. IPE INSIGHTS, IMPROVEMENTS AND COMMITMENTS , . . . . . . . . . . 12

5. REFERENCES ....................................13 APPENDIX A: IPE EVALUATION AND DATA

SUMMARY

SHEET . . . . . . . A- 1 l 1

Hatch IPE Back-End Review ii December 1994

EXECUTIVE

SUMMARY

L l This is a review of the back-end portion of Georgia Power Company's Individual Plant Examination (IPE) of the Edwin I. Hatch Nuclear Plant, Units I and 2.2 The two units are nearly identical BWRs with Mark I containments, and the analyses show very few I differences. The back-end analysis team concentrated on Unit 2, but performed complete analyses of Unit I as well. Because of the similarities, all sensitivity studies were done for Unit 2 only.

The Hatch IPE was performed based on level I and Level II probabilistic risk assessments (PRAs) that applied well-established practices within the PRA community, practices that have been used for a very large number of full-scope PRAs. The IPE team consisted ofin-house staff and consultants from PLG, Inc., and Westinghouse /Fauske & Associates, Inc.

Independent in-house personnel and an independent consultant from General Electric Company reviewed the IPE submittal.

The calculated total core damage frequencies are 2.2E-5 per year for Unit I and 2.4E-5 per year for Unit 2, which are both about 5 times higher than that for Peach Bottom, which was the NUREG-1150 BWR/ Mark I plant (4.5E-6 per year). For Unit I and Unit 2, the conditional probabilities of containment failure, given core damage, are found to be about 24%/21 % for early failures, 20%/20% for late failures, and 1%/1 % for containment bypass sequences. The probability of no failure is calculated to be 49%/53% while containment integrity maintained by venting is calculated to be 5 %/5 %.

The analysis team developed containment event trees to analyze the containment performance under severe accidents. The team used the MAAP computer program to help determine plant damage state behavior (success or failure), accident timing, radionuclide release fractions, and radionuclide decontamination factors.

The criteria used to determine if a " vulnerability" exists at either unit at Hatch were: 1) any accident class with frequency > IE-4/ year or that contributes more than 50% of the total core damage frequency (CDF),2) containment-bypass sequences totaling > IE-5/ year or contributing more than 20% to the total CDF, and 3) any source-term analysis bin representing containment failure or impairment with CDF > IE-5/ year, and in which a single function, system, operator action, or other element can be identified that substantially contributes to the CDF. The analysis team found no significant vulnerabilities associated with either of the two Hatch units.

The team found that two functional-sequence groups are the dominant contributors to large releases resulting from severe accidents. In ong group, vessel depressurization has failed along with drywell sprays; no injection is available; and early containment failure (relative to l

1

' Georgia Power Company, " Plant Hatch Units I and 2:

Generic Letter 88-20 Response, Individual Plant l

I Examination Submittal *, December 1992 lii D m = W 1994 Hatch IPE Back-End Review

L F .. . .. .

vessel failure) is in the drywell, bypassing the torus and its fission-product-removal capabilities. In the other group, containment heat removal is lost, and containment failure RPV failure occurs, natural

. occurs in the drywell before RPV failure occurs; ht us when fission-product-removal mechanisms cannot mitigate the release.

i The analysis team responded to the CPI program, recommendations as follows: 1) alternate water supply systems for drywell spray / vessel injection are available at Hatch'and EOPs address using these systems, 2) a reliable system is available at Hatch to depressurize the vessel, thus vessel depressurization is not a concern, and 3) the Revision 4 version of the BWR Owners Group EPGs has been implemented at Hatch, including many operator actions I appropriate for managing severe accidents.

'Ihe submittal notes that the hardened vent, installed in response to NRC GL 89-16, provides an exhaust line from the torus and drywell around the Standby Gas Treatment System to the plant stack, to mitigate core-damage sequences resulting from long-term loss of decay-heat removal. Besides the hardened vent, no other back-end plant improvements were reported in the submittal.

I i

iv Decernber 1994 Hatch IPE Back-End Review

I

1. INTRODUCTION i This technical evaluation mport (TER) documents the results of ~ ,: view of the Edwin I.

Hatch Nuclear Plant individual plant examination (IPE) back-end submittal for Hatch Units 1 and 2. This technical evaluation report complies with the requirements for submittal-only reviews of the U.S. Nuclear Regulatory Commission contractor task order, and adopts the NRC review objectives, which include the following:

e To determine if the IPE submittal provides the level of detail requested in the

" Submittal Guidance Document," NUREG-1335 e To assess the strengths and the weaknesses of the IPE submittal e To complete the IPE Evaluation Data Summary Sheet

• To pose a preliminary list of questions about the IPE submittal, based on this limited review.

Section 2 of the TER summarizes our findings and briefly describes the Hatch IPE submittal as it pertains to the work requirements outlined in the contractor task order. Each portion of Section 2.1 corresponds to a specific work requirement. Section 2.2 sets out our assessment of the Hatch submittal's strengths and weaknesses. Section 3 presents our overall evaluation of the Hatch IPE submittal, based on our submittal-only review. Section 4 outlines the IPE insights, plant impmvements identified, and the utility commitments. Appendix A presents an evaluation summary sheet completed on the Hatch IPE.

I Hatch IPE Back-End Review 1 December 1994

' 2. CONTRACTOR REVIEW FINDINGS 2.1 Review and Identification of IPE Insights

~

This section is structured in accordance with Task Order Subtask 1. -i 1

- 2.1.1 General Review of IPE Back-End. Analytical Process j 2.1.1.1 Completeness .

The Hatch IPE back-end submittal is essentially complete with respect to the level of detail i

'I requested in NUREG-1335. He submittal appears to meet the NRC sequence selection screening criteria described in Generic letter 88-20.

I 2.1.1.2 Description, Justification, and Consistency .

1 The IPE methodology used is described clearly. He approach followed is consistent with j the basic tenets of Generic I.etter 88-20.

2.1.1.3 Process Used for IPE As noted in Section 2.3, page 2.3-1 of the submittal, for I.evel I the analysis followed the PRA methodology developed by PLG, which is commonly referred to as the linked-event-tree methodology. His approach has the advantage that it eliminate < the need for support-state binning, and allows an integrated analysis and quantification of the containment response. He Level I model quantification led to identification of 11 classes / subclasses of accident sequences, all of which were analyzed using a single larEe Containment Event Tree (CET). For the I.evel II analysis, the MAAP computer code was used to calculate severe i accident event timing and containment loads for representative sequences. ,

2.1.1.4 Peer Review of IPE The Hatch IPE analysis team consisted of in-house staff and consultants from PLG, Inc. and Westinghouse /Fauske & Associates, Inc. (FAI), with FAI taking the lead for the I.evel H -

portion of the IPE. IPnt in-house personnel and an independent consultant from General Electric Company peer-reviewed the IPE submittal. As noted in Section 5.2 of the submittal, peer review team members had numerous meetings with the analysis team through the duration of the project, and commented on various individual issues in real time.

Section 5.3 of the submittal discusses the peer review team's comments and resolutions.

De discussion of the peer-review process is extensive and provides insights into how the process worked. It is clear that the peer-review process had a positive impact on the overall IPE analysis, and it appears that its structure and modus operandi were satisfactory. In particular, the text in Section 5.3 indicates that considerable attention was paid to the level II issues.

Hatch IPE Back-End Review 2 Demmber 1994

-<w ,,n. ,, , - - - --c...,, . , . .~-r , 7._ .. - - - , .

2.1.2 Containment Analysis / Characterization.

2.1.2.1 Front end Back-end Dependencies i l

The IPE team coupled the frontamd analysis to the back-end analysis by binnirig the l front-end dominant sequences into a'few groups with similar back-end characteristics. The back-end analysis was performed in two phases. Ihe first phase involved torting and gmuping the level I cutsets from the dominant accident sequences into five classes (I, II, III, .;

IV, V) based on functional similarities. Classes I and III were further subdivided into i Subclasses IA, IB, IC, and ID, and similarly for IIIA, IIIB, IIIC, and IIID. This led to a  !

1 total of 11 classes / subclasses in all. Through the sorting pmcess, these Il classes and subclasses were made mutually exclusive, although as defined initially they were not -!

w=ily so. Every class / subclass was associated with a large number of Ievel I

- individual accident sequences, but they were gmuped together for the purpose of Ievel II analysis.

The second phase involved using the Containment Event Tree (CET) to sort the various ,

fmnt-end accident sequences into back-end "end states," each with a unique identifier.

Dozens of these "end-states" emerged from the single CET that was used in the analysis.

Each end-state was then assigned both a plant damage state (PDS) and a release mode (RM),  ;

which comprised a smaller number than the number of end-states - that is, more than one i end-state can map into a single PDS or into a particular RM.

The PDS binning differentiates sequences by the status of systems and equipment important ,

for the source-term analysis. The PDS categorization system has three designators for

" reactor status," seven for " containment status," and three for " status of debris cooling."

"Ihe RMs are used to differentiate sequences based on the magnitude and timing of radiological releases, and they are broadly grouped into six magnitude categories and four i timing categories. The magnitude categories, many of which have subcategories as well, l are:

N intact containment / reactor vessel S containment failure occurs, release with torus scrubbing E early containment failure, preceding or coincident with reactor vessel failure, torus bypassed L late containment failure, torus bypassed I failure to isolate containment prior to core damage B containment bypassed directly.

I The four timing categories are: j N no release, except normal leakage E carly, during the first 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> of the event I intermediate, during the first 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> (overlaps release mode E)

L late, on the order of 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> or more.

Hatch IPE Back-End Review 3 December 1994

- - . . - - - . - , _ _ , , - , - - - . . + . _ _ - _ _ _ _ _ _ _ , . . _ _ . - - - _ , - - , . . ~ - , _ . . - . - -

i ,: - .- r

'Ihe Hatch CET contains the following nine questions, which were used to differentiate the l various back-end sequences and to sort the events. (Figure 4.5-1 of the submittal) e What is the CET accident class / subclass? .

e Is the containment-isolation function successful after core damage? ,

e What is the containment status? ('Ihere are three possibilities besides containment successfully isolated: torus gas-space failure, drywell failure, or drywell shell failure.) ,

e Is the reactor vessel at high pressure or depressurized after core damage and before  !

vessel failure? l e Is injection available after depressurization of the vessel and before vessel failure, for the purposes of in-vessel recovery of the core <1amage accident? (The IPE analysis takes no credit for this injection as a means of in-vessel recovery.)

e Are drywell sprays available after core damage?

e Is low-pressure injection available following vessel failure?

e Is containment heat removal (torus cooling) available following core damage?

e What is the status of the hardened vent after core damage? 'Ihe three possibilities are successful vent from the torus, successful vent from the drywell, and failure of the vent from both locations. l We conclude that the Hatch IPE team conveyed the important vessel and plant equipment conditions from the front-end analysis to the back-end analysis.

2.1.2.2 Sequences with Significant Probability i I

In the back-end analysis, the CET led to 22 end states with frequencies greater than IE-9/ year (see Tables 4.6.1 and 4.6.2 of the submittal). Of these, only five have frequencies that are greater than IE-6/ year. We conclude that the submittal meets the sequence selection criteria, as outlined in Appendix 2 to Generic Letter 88-20.  ;

2.1.2.3 Failure Modes and Timing Failure Modes Section 4.4 of the submittr.1 chapcterizes 10 different modes of Hatch containment failure and associated phenomenology. The 10 examined were:

o Containment overpressurization e Direct containment bypass e Failure to isolate containment e Liner melt-through e Steam explosions Hatch IPE Back-End Review 4 Deceznber 1994

g cm, .

I v

.e Direct containment heating l e- Deflagration and detonation of hydrugen e Molten core-concrete interacdons leading to late containment failure i e Brust forces at reactor vessel failure . j e Thermal attack on containment penetrations. ,

All 10 failure modes and associated phomenology are discussed thoroughly in separate '

subsections under Section 4.4, with sufficient detail to allow a reader to understand their potential for Hatch. Of these 10, only the first four were found to be important enough to 'l play a role in the Level H analysis. The way that the other six are discussed and the basis for the conclusions that they are not important seem to be convincing.

l De discussion of containment strength against overpressurization (Section 4.4.5) indicates that the potentially important failure modes are failure of the drywell shell (mean failure  ;

pressure of 128 psig), of the drywell head closure (122 psig), and of the vent-line bellows in the torus gas space (100 psig). nree different fragility curves were developed for these three failure modes: when combined into a total fragility curve, the median failure pressure  ;

~

was found to be 98 psig). The asymptotic failure probability values are as follows: vent line bellows in torus gas space (79 %), drywell head closure (21 %), and drywell shell  ;

(negligible). He failure size depends on location: for the torus air space it was 7 ft2 , while for the drywell head it was 0.25 ft2 for non-ATWS sequences and 0.97 ft2 for ATWS l sequences.  ;

l The Hatch containments will also fail even at low pressure due to elevated temperatures: at 500T the drywell head is eWM to fail at about 60 psig; at 900T it is likely to fail at any pressure. There are slight differences between Units 1 and 2 for this type of failure due to slight differences b design criteria. j Timine of Containment Failure: ne MAAP computer code was used to calculate possible times of failure for the various failure modes under predicted containment loads.

As discussed below in Section 2.1.3.2 of this TER, the IPE team found that the dominant contributor to large early containment release was a sequence with early drywell failure m j which the torus was bypassed (see Table 4.8-1 of the submittal). l The Hatch IPE team appears to have addressed containment failure modes and timing in sufficient detail.

2.1.2.4 Containment Isolation Failure j l

Section 4.4.2, page 4.4-2 of the submittal notes khat the analysis emphasized those l penetrations and pathways which, if open, would prevent the containmect from pressunzmg.

He isolation rystem consists of both the penetration piping and the contrul systems. As shown in Table 4.6.3,0.2% of the level-I core-damage frequency is associated with failure to isolate containment in both units at Hatch, ne analysis seems to have be:n done according to the guidelines in GL 88-20.

i

't Hatch IPE Back-End Review 5 December 1994 l

., ~ -. - - - - .- . . .

l 3 .-.4 j b

2.1.2.5 System / Human Responses As noted in Section 2.1.1.3 of this report, for Level I the analysis followed the PRA '

methodology developed by PLG, which is commonly referred to as the linked-event-tree '

methodology. This appmach has the advantage that it elim* mates the need for support-state

+ binning, and allows an integrated analysis and quantification of the containment response.

j j

Although the licensee provided much data in Section 3, the indapandant review of post-core damage system and human responses was limited due to discussions in the submittal that j

were less than complete.

2.1.2.6- Radionuclide Release Characterization ne Hatch IPE team characterized radioactive releases by running the MAAP code for 14 sequences. Dese 14 sequences were selected so that every one of the 11 accident classes ~;

and subclasses was represented; so that each type of containment status and each debris-bed-cooling status was represented if it had a frequency > IE-9/ year; and so that one bypass sequence and one containment-isolation-failure sequence were represented. He selection process was intended to satisfy the screening criteria in Section 2.1.6 of -

NUREG-1335, and it appears that this was accomplished successfully. De analysis team' i

determined that the same 14 sequences are ny,satative of both Unit I and Unit 2, whose slight differences made almost no difference to the analyses or the insights. (Almost the only. ,

difference between the units that affects the level H analysis is that Unit 2 has a separate chiller for the containment fan coils, which significantly increases the probability of Unit 2 1

losing all of its drywell cooling, as discussed in Table 3.4.3 of the submittal.)

\

For purposes of dealing with the releases, the various fission products were grouped into -

three broad categories (see Table 4.7.1 of the submittal): noble gases (Xe, Kr); volatile species (CsI, CsOH, RbI, TeO 2, Te2), and non-volatiles (the remainder).

The final binning of the level H results placed all releases into four " Release Categories" (see Table 4.7.2 of the submittal), as follows:

Percentane of Radiorumlidae Released Relane Catenorv Nobles Volatiles Non-vointitee A i 100 % < 0.1 % and < 0.1 %

B i 100 % 0.1 to 1% and < 0.1 %

C i 100 % 1 to 10% or 0.1% but <2%

D 1 100 % > 10% or > 2% i Release Category D clearly represents the most severe releases. He dominant contributors to Category D are discussed below in Section 2.1.3.2 of this TER.

1 I

i 6 December 1994  ;

Hatch IPE Back-End Review )

_ m . , . _. .. .

'i .

l 2.1.3 Accident Progression and Containment Perfonnance Analysis l

2.1.3.1 Severe Accident Progression l

Section 4.4 of the submittal discusses the containment failure modes, and Section 4.6 l describes the accideut-progression analysis. The IPE team used the MAAP Version 3.0B computer prognm to determine accident timing, radionuclide release fmetions, and radionuclide decontamination factors. Although the submittal provided numerous tables and -

figures displaying the results of the MAAP analysis, the Ladm-ht review of the

. assumptions was limited due to detail in the submittal that was less than complete. l 2.1.3.2 Dominant Contributors: Consistency with IPE Insights Table 1 in this report shows the results of a comparison of the darainaat contributors to the' Hatch Unit I containment failure modes with those contributors identified during IPEs performed at the Fitzpatrick, Oyster Creek, Browns Ferry, Duane Arnold, and Dresden plants, and with the NUREG/CR-1150 PRA results obtained at Peach Bottom. In Table 1 on page 14 of this report, both early and failure-to-isolate delayed releases at Hatch are i combined and shown under the early release group; the intact-containment category includes the no-vessel-breach category and the torus-venting category; the torus-sembbing case is not shown because there are no sequences (<0.1 %) in that category.

No major differences exist among the results for the various plants as shown in Table 1.

According to the table, the probability of no failure, i.e., containment intact, is somewhat higher at Hatch Unit I than at the other plants listed. However, there is no important i significance attached to the numerical differences displayed in Table 1 because they are smaller than the accumcy with which they have been calculated, j

1 As described in Section 2.1.2.1 of this report, the IPE analysis team categorized possible 1 radionuclide releases at Hatch according to both Release Mode and Timing categories. The types of release and the fraction of total core-damage frequency represented by each are

{

shown in Table 2 on page 15 of this mport. De release-timing information is also shown in Table 2. One principal insight evident from this table is that early releases represent 2%/3 %

of total core damage frequency for Units 1 and 2, respectively, while sequences with late releases account for 29%/25 %, and containment bypass sequences account for 1 %/1 %. The probability of no failure is calculated to be 49%/53 % and containment integrity maintained by venting is calculated to be 5%/5%.

Four functional accident sequence groups with frequencies above IE-9/ year are the dominant contributors to Release Category D, the category in which the release of volatiles exceeds 10%. He total core-damage frequency for Release Category D is 23% for Unit 1 (22% for Unit 2) of the total core-damage frequency for all sequences. Dese four functional sequence groups (see Table 1.4-1 of the submittal) and their annual frequencies are as follows:

Functional Secuence Grono 1: (Unit 1: 3.5E-6; Unit 2: 3.7E-6): These are sequences in  !

which vessel depressurization has failed along with drywell sprays. No injection is available, Hatch IPE Back-End Review 7 Demnher 1994 1

a , Am p4 .q, n ._#4 a. 1 A,._ Asa 4 #_+-1 4 - A u =e- - +4 e ,

and early containment failure (relative to vessel failure) is in the drywell, bypassing the torus and its fission-product-removal capabilities.

Functional Seouence Group 2: (Unit 1: 1.0E-6; Unit 2: 8.8E-7): These are sequences in which containment heat removal is lost, and containment failure occurs in the drywell before RPV failure occurs. Hus when RPV failure occurs, natural mechanisms for fission-product removal cannot mitigate the release. Also, ECCS systems are conservatively assumed to

- have failed because of harsh conditions in the containment after containment failure.

Functional Seauence Gmup 3: (Unit 1: 1.7E-7; Unit 2: 1.8E-7): These am sequences with a containment bypass, and a very large (77%) release of volatile fission products.

Functional Sequence Groun 4: (Unit 1: 5.8E-8; Unit 2: 7.2E-8): nese am sequences I involving ATWS events, and containment failure occurs in the drywell. Also, ECCS systems )

- are conservatively assumed to have failed because of harsh conditions in the containment i after containment failure.

Only the first two groups above exceed the IPE screening criterion of 10% or more volatile fission-product release and a frequency above IE-6/ year. The analysis team used this j screening criterion to determine whether an evaluation of possible procedural or hardware modifications should be considered.

2.1.3.3 Characterization of Containment Performance Section 4.6.3 of the submittal describes the percentage of core-damage frequency represented by the various possible containment failure modes. The findings (Table 4.6.3 of the submittal) are shown in Table 3 on page 16 of this report.

The IPE team appears to have characterized the containment performance in adequate detail.

2.1.3.4 Impact on Equipment Behavior '

l De Hatch IPE submittal does not appear to describe the survivability of equipment under I severe accident conditions in enough detail to allow for a review. Certain assumptions have been made about some equipment, most notably (i) that ECCS equipment in the containment  ;

cannot function in sequences in which containment failure precedes core damage and vessel failure and (ii) that no credit is taken for reactor-building systems after containment failure. l However, the principal accident sequences are not described completely.  ;

In Section 4.7.3 of the submittal, the individual; accident sequences analyzed using MAAP are discussed in detail. For each sequence, the discussion covers which equipment items are available or failed, and if any human errors are embedded in a given sequence. He licensee l did not provide the degree of completeness to i_a+peMantly reviety the survivability of equipment, including failure probabilities and dependencies. j 1

I Hatch IPE Back-End Review 8 Ihmher 1994 I

l l

1

?

2.1.4 - Reducing Probability of Core Damage or Fission Product Release 2.1.4.1 Definition of Vulnerability l

In Section 1.5, page 1.5-1 of the submittal, it is noted that:

The Hatch IPE was performed consistent with GL 88-20 and did l not identify any plant-specific vulnerabilities to severe accidents. i Section 3.4.2.1 cites three criteria that were used to determine if a vulnerability exists at i either unit at Hatch

• i e Any accident class with frequency > IE-4/ year or that contributes more than 50% of  ;

the total core-damage frequency (CDF) i e Containment-bypass sequences totaling > IE-5/ year or contributing more than 20% .

to the total CDF  !

e Any source-term analysis bin representing containment failure or impairment with CDF > IE-5/ year, and in which a single function, system, operator action, or other ,

element can be identified that substantially contributes to the CDF.

2.1.4.2 Plant Improvements The analysis team reviewed the IPE results to identify any " cost-effective modifications" that would significantly reduce the CDF or reduce releases. None were identified.

The criteria used in this review (Section 3.4.2.1) were:

e Any accident class or subclass with CDF > IE-6/ year.

e Containment bypass sequences totaling > IE-7/ year.

e Any source-term analysis bin with CDF > IE-6/ year, and with a release exceeding 10% of the volatile fission products.

Only one I.evel-II-type modification (the hardened vent) was identified that fits the above criteria. It is discussed below. However, the submittal notes in Section 6.2 l that the Hatch IPE analysis took about 4 years to complete, and " physical and i procedural modifications for previously identified issues were developed and scheduled for implementation without waiting for the completion of a formal evaluation. The Hatch IPE was also affected by certain modifications driven by other

. rquiatory issues."

The hardened vent, installed in response to NRC GL 89-16, is discussed in Section 3.2.1.27

, of the submittal. Its installation was coordinated with the IPE analysis "to ensure the design adequately addressed loss-of-decay-heat-removal sequences with respect to available support Hatch IPE Back-End Review 9 Wher 1994  !

e - . --- - + . . . - - , _ - _ y < -- -

y w - -

i i

systems" (Section 6.2 of the submittal). It provides an exhaust line from the torus and i drywell around the Standby Gas Treatment System to the plant stack to mitigate core-damage sequences resulting from long-term loss of decay-heat removal. '

Besides the hardened vent, no other back-end plant improvements were reported in the i submittal.

2.1.5 Responses to CPI Program Recommendations Generic Letter No. 88-20, Supplement No.1, reiterates the following recommendations made by the Containment Perfonnance Improvement'(CPI) Program pertaining to Mark I  ;

containments: l e Alternate water supply for drywell spray / vessel injection o Enhance reactor pressure vessel depressurization system reliability by upgrading the  ;

power supply and the pneumatic supply to the SRVs (safety / relief valves) l

• Implement emergency procedures and training (BWROG EPGs Rev 4). l Supplement No. I also notes that the above improvements should be considered in addition to  ;

improvements that stem from the evaluation and implementation of hardened vents.  ;

De Hatch response to the CPI Program recommendations is found in Section 6.4.1 of the submittal. A summary of the submittal's discussion is as follows: )

i e The IPE analysis did not take credit for alternate water-supply systems, although j Hatch has EOPs that address the alignment and operation of various alternate systems, )

such as injection to the vessel through the LPCI injection paths, or to the drywell l through the RHR spray header using RHR service water or fire protection systems. l e No enhancements are noted concerning the reliability of vessel depressurization capability at Hatch.

e De IPE submittal notes that the Revision 4 version of the BWR Owners Group EPGs l has been implemented at Hatch, including many operator actions appropriate for j managing severe accidents.  ;

1 2.2 IPE Strengths and Weaknesses ;

2.2.1 IPE Strengths j

1. De Hatch IPE team performed containment analysis using a single CET with 11 different classes / subclasses of accident sequences brought forward from the Level I PRA. This is an adequate way to carry out the back-end analysis.

Hatch IPE Back-End Review 10 December 1994

,~.a - - a +

)

i

2. The IPE' team appears to have conveyed the important vessel and plant equipment conditions from the front-end analysis'to the back-end analysis.

3. Extensive information on various containment failure modes was provided in the _

submittal.

I The IPE team addressed the various release modes, timing, and containment failure l 4.

l modes in a thorough manner.

l

5. The sensitivity analysis done using MAAP seems to have covered the important issues thoroughly. I I

6. An =d-=!=** reg-:=m to the Containment Performance Improvement (CPI) Program recommendations appears to have been made in the submittal.

j

7. An Ladmandant in-house review did take place, and the information provided in the submittal is sufficient to judge that both the breadth and depth of the back-end portion of the review process were satisfactory.

i 2.2.2 IPE Weaknesses 1 Although the licensee has provided much information, the level of detail is less than l

1. i complete in three areas. His has limited the Ladmandent review of the post-core damage system and human responses, accident progression assumptions, and equipment survivability (including failure probabilities and hy tries).

3. OVERAIL EVALUATION I

As discussed in Section 2 of this TER, the IPE submittal for Hatch contains a large amount of back-end information,' which contributes to the resolution of severe accident vulnerability issues for both Hatch units.

The key points of our technical evaluation of the Hatch back-end submittal are summarized as follows:

e Despite the few limitations discussed above in Section 2.2.2, the Hatch IPE submittal demonstrates a good understanding of the impact of severe accidents on containment failures and subsequent radionuclide releases, and it does provide the level of detail i i

requested in NUREG-1335.

e Having evaluated the key large-release accident sequences in terms of the potential for improvements to reduce either the frequencies or consequences of these sequences, the IPE team concluded that no cost-effective improvements exist. . De team's evaluation demonstrates a thorough understanding of the IPE program process and purpose.

11 Damher 1994 Hatch IPE Back-End Review  !

o .

4. IPE INSIGHTS, BiPROVEMENTS AND COhmiITMENTS As noted in the submittal, the Hatch IPE team calculated a 23 % probability of a large release

(> 10%) of volatile radionuclides to the environment, given a core-damage event at Unit 1.

For Unit 2, this fraction is 22%. The team gained a number of insights into how the consequences of a core damage event at Hatch could be mitigated, including the following (Section 6.1 of the submittal):

e ' The hardened vent provides a means of removing heat from the containment that is independent of the RHR Service Water pumps and Plant Service Water pumps that are located in the intake stmeture.

e The current EOPs (Revision 4 of the BWROG EPGs) are structured to allow the operators maximum time for recovery of high-pressure injection.

e The fact that the primary containment is inerted with nitrogen during operation prevents hydmgen deflagration and detonation.

e Toms cooling and drywell sprays following RPV failure help significantly to avoid containment failure during many core-damage sequences.

e In the event of loss of torus cooling, venting is important in limiting offsite releases.

e Containment failure fmm overpressure is delayed longer at Hatch than at many otLer BWR-Mark I plants, because the rated thermal power is smaller than at many otherwise similar plants.

e Although sequences with a bypass of containment can release a very large fraction (77%) of the volatile fission pmducts, they comprise only about 1% of all core-damage sequences at Hatch.

e Sequences involving a failure to isolate containment are a very small fraction (0.2%) of all core-damage sequences at Hatch.

The IPE submi:.tal potes the following back-end-type plant improvements, neither of which was made explicitly because of the IPE, but each of which is an important safety improvement relevant to back-end issues:

e The installation of the hardened vent. :

e The incorporation of the improved EOPs, which is Revision 4 of the BWROG EPGs.

Acconting to the submittal, Hatch has made no commitments to carry out other plant or procedural changes as a result of the back-end analysis. l Hatch IPE Back-End Review 12 December 1994

5. REFERENCES 9

1. Georgia Power Company, " Hatch Units 1 and 2: Generic Letter 88-20 Response, Individual Plant Examination Submittal," 1992.

2. U.S. Nuclear Regulaton Commission, Generic letter 88-20, " Individual Plant l Examination for Severe Accident Vulnerabilities - 10 CFR 50.54(f)," 1988.

3. U.S. Nuclear Regulatory Commission, Report NUREG-1335, " Individual Plant Examination: Submittal Guidance," 1989.

4. U.S. Nuclear Regulatory Commission, Report NUREG-1150, " Severe Accidents Risks: f An Assessment for Five U.S. Nuclear Power Plants," 1989.

t Hatch IPE Back-End Review 13 December 1994

Table 1 Containment Failure as a Percentage of CDF:

Hatch (Unit 1) Results Compared with the Results of the Fitzpatrick, Oyster Creek, Biswns Ferry, Duane Arnold, and Dresden IPEs and with the Peach Bottom NUREG-1150 PRA Results I Containment Fitzpatrick Oyster Browns Duane Dresden Peach Hatch Failure IPE Creek Ferry Arnold IPE Bottom / Unit 1 IPE IPE IPE NUREG IPE

-1150 CDF 1.9E-6 3.2E-6 4.8E-5 7.8E-6 1.8E-5 4.5E-6 2.23E-5 (per year)

Early failure 60 16 46 47 3 56 21*

Bypass na 7 na 0 0 na 1 late failure 26 26 26 32 86 16 29*

Intact 3 0 3 21 11 18 49 No vessel 11 51 25 na na 10 na breach

• includes torus venting i

Hatch IPE Back-End Review 14 Dexmber 1994

M Table 2 Types of Releases and Timing of Releases: '

Percentage of Core Damage Frequency Feiwnare of Core-Damage Fmouency [

Tyne of Release: Unit 1 Unit 2  ;

i No release beyond nonnal leakage 49 % 53 %

I. ate containment failure mode 20 % 20 %

. Early containment failure mode 24 % 21 %

Torus venting 5% 5% ,

Containment bypassed 1% 1%

Containment isolation failure 0.2 % 0.2 %

Timine of Release:

i No release beyond normal leakage 49 % 53 % '

Late release 29 % 25 %

Intennediate release 20 % 20 %

Early release 2% 3%

1 l

1 i

l l

Hatch IPE Back-End Review 15 D=her 1994 l

i Table 3 Containment Status and Reactor Status:

Percentage of Core-Damage Fnquency .

Percentmoe of Core-Dammae Freauency Containment Status: Unit 1 Unit 2 Containment intact 49 % 53 %

Overtemperature failure 20 % 20 %

Overpressure failure 25 % 21 %

Torus venting 5% 5%

Containment bypassed 1% 1%

Containment isolation failure 0.2 % 0.2 %

,. Drywell venting < 0.01 % < 0.01 %

Reactor Status:

Vessel failure at high pressure 76 % 80 %

Vessel failure at low pressure 24 % 20 %

Arresthecovery in vessel (assumed to be zero, not analyzed to be zero) l 1

I I

i Hatch IPE Back-End Review 16 December 1994

. . . l APPENDIX A 1 IPE EVALUATION AND DATA

SUMMARY

SHEET ,

BWR Back-end Facts j Plant Name l Edwin I. Hatch, Units 1 and 2 Containment Type )

l Mark I Unique Containment Featuris None found Unique Vessel Features None found Number of Plant Damage States 11 " classes" or " subclasses" were used, similar to PDSs Ultimate Containment Failure Pressure 98 psig median (both Units)

Additional Radionuclide Transport And Retention Structures Reactor building retention was not credited after containment failure Conditional Probability 'Ihat The Containment Is Not Isolated 1

0.2% (both Units)

Important Insights, including Unique Safety Features  ;

None identified

'i4 Hatch IPE Back-End Review A-1 December 1994

. .= .

APPENDIX A (contbved)

IPE EVALUATION AND DATA

SUMMARY

SHEET I

l Implemented Plant Improvements t l- None identified l

C-Matrix Not enough information presented in the Hatch IPE back-end submittal to develop a C-Matrix I

t A-2 December 1994 Hatch IPE Back-End Review

7 -

1 0

L l[

L HATCH UNITS 1 AND 2 INDIVIDUAL PLANT EXAMINATION TECHNICAL EVALUATION REPORT (HUMAN RELIABILITY ANALYSIS) l I

l Enclosure 4

CONCORD ASSOCIATES,INC. cuTR 94-019-027 Systems Performance Engineers EDWIN I. HATCH NUCLEAR POWER PLANT UNITS 1 AND 2 TECHNICAL EVALUATION REPORT ON THE INDIVIDUAL PLANT EXAADNATION HUMAN RELIABILITY ANALYSIS l FINAL REPORT By P.M. Haas PJ. Swanson l

Prepared for i

U.S. Nuclear Regulatory Commission Omce of Nuclear Regulatory Research  !

Division of Safety Issue Resolution Draft Repon May,1994 Final Report Noveiaber,1994

. . _ ~ . . . . . . . - . . .- . _ - . . . - . . . - - - _ _ . _ . .. . . - - . = _ . . . . - . . . _ . . _ . _ _ _

11915 Cheviot Dr. 725 Pellissippi Parkway 6201 Picketts Lake Dr.

Herndon, VA 22070 Knoxville,TN 37932 Ackworth, GA 30101 (703) 318-9262 (615) 675-0930 (4N) 917-0690

..